Neural correlates of change in major depressive disorder anhedonia following open-label ketamine

Anhedonia—reduced enjoyment or desire for pleasurable activities—is a core symptom of major depressive disorder (MDD), affecting roughly 40% of patients and predicting greater illness severity, poorer treatment response and elevated short-term suicide risk. The authors note that standard antidepressants often have limited efficacy for anhedonia and may even produce emotional blunting; no medication is currently approved specifically to treat this symptom. Preclinical and early clinical work implicates glutamatergic dysfunction and rapid-acting NMDA receptor antagonists as promising avenues, and previous work in bipolar disorder suggested a specific anti-anhedonic effect of ketamine linked to dorsal anterior cingulate cortex (dACC) metabolism. Lally and colleagues therefore set out to examine whether a single open-label subanaesthetic ketamine infusion reduces anhedonia in medication-free, treatment-refractory MDD patients, and to explore neural correlates of any change using [18F]-FDG PET in a subsample. The protocol also included randomisation, begun 4–6 hours post-infusion, to adjunctive oral riluzole versus placebo for 28 days; the present paper focuses on exploratory analyses of anhedonia measured with the Snaith–Hamilton Pleasure Scale (SHAPS) and on PET measures of regional cerebral glucose metabolism associated with change in anhedonia.

The study recruited 52 inpatients with treatment-refractory MDD (non-psychotic), defined as failure of two or more adequate antidepressant trials, and currently in a major depressive episode with MADRS ≥ 22. Diagnoses were established via the Structured Clinical Interview for DSM-IV and unstructured clinical interview. Exclusion criteria included recent substance dependence, prior ketamine/riluzole/phencyclidine use, recent ECT, recent fluoxetine, significant unstable medical conditions, and pregnancy. All participants were medication-free following a 2-week taper (5 weeks for fluoxetine) before infusion. Patients received a single open-label intravenous infusion of ketamine hydrochloride 0.5 mg/kg over 40 minutes. Between 4 and 6 hours after infusion, participants were randomised (double-blind to riluzole/placebo) to receive oral riluzole or matched placebo twice daily for 28 days (initial 100 mg/day titrated weekly by 50 mg to a maximum of 200 mg/day, with reductions allowed for side effects). No other CNS-active medications or psychotherapy were permitted. Symptom ratings were obtained at 60 minutes pre-infusion (baseline), 40, 80, 120 and 230 minutes post-infusion, then daily for 28 days. The primary outcome for the original trial was the MADRS; this analysis centres on the SHAPS, a 14-item self-report measure of anticipatory anhedonia (items scored 1–4, range 14–56), with clinically significant anhedonia defined as disagreement (score 3 or 4) on at least three items. A subsample of 20 patients (44%) underwent FDG-PET: one scan at baseline (1–3 days pre-infusion) and a second beginning ~2 hours and ending ~3.5 hours post-infusion, prior to initiation of riluzole/placebo. PET acquisition used a GE Advance scanner with dynamic cardiac imaging and venous blood sampling to generate parametric regional cerebral metabolic rate of glucose (rCMRGlu) images; anatomical MRI was acquired for registration. PET analyses included region-of-interest (ROI) analyses focused on ventral striatum (VS) and orbitofrontal cortex (OFC) and whole-brain voxel-wise analyses using SPM5. For change analyses, the percentage change in SHAPS from baseline to 230 minutes post-infusion ([(ketamine – baseline) / baseline]) was used as the regressor; PET difference images were created by subtracting baseline from post-ketamine scans and normalising by global mean. Small-volume correction (10 mm sphere) was applied to test the previously implicated dACC coordinate. Statistical analyses used two-tailed tests with significance at p < .05 (p < .1 as trend) and were performed primarily in SPSS 21. Longitudinal symptom change was analysed with linear mixed models (restricted maximum likelihood) using an autoregressive moving averages covariance structure; post-hoc simple effects were Bonferroni-corrected for 32 time points. PET correlations used Pearson and partial correlations for ROIs, and SPM5 whole-brain regressions with an initial voxelwise threshold p < .001 uncorrected and cluster-level family-wise error correction (cluster-level FWE p < .05, p < .1 considered trend). The imaging analyses were explicitly exploratory and no correction for multiple comparisons was applied across all tests.

Sample characteristics and psychometrics: Fifty-two medication-free, treatment-refractory MDD inpatients were included; 87% met the predefined SHAPS threshold for clinically significant anhedonia at baseline. Total SHAPS correlated with MADRS at baseline (r(52) = .47, p < .001). Comparison of MADRS and MADRS anhedonia item between consensus pre-taper ratings and the baseline time point showed no significant change, suggesting the drug-free/taper period did not significantly alter depression or anhedonia measures prior to ketamine. Dissociation (CADSS) scores were significantly lower at 230 minutes than at baseline (t(51) = 3.74, p < .001), and 230-minute CADSS did not differ from 24-hour scores. Primary clinical outcomes: Lally and colleagues found no evidence that riluzole augmented ketamine's anti-anhedonic effect: there was a significant effect of time on SHAPS across days 1–28 (F(27,610) = 1.910, p < .004) but no main effect of drug (riluzole versus placebo) (F(1,51) = 0.201, p = .656) and no time-by-drug interaction (F(27,610) = 1.204, p = .221). Given this, subsequent models pooled groups and tested change over time following the single ketamine infusion. The time-only model showed a significant effect (F(32,642) = 3.355, p < .001), indicating SHAPS scores decreased across the 28-day follow-up. Bonferroni-corrected post-hoc comparisons versus baseline indicated significant or trend-level reductions at 40 minutes (t(1010) = 8.57, p_corr < .001), 80 minutes (t(1312) = 4.55, p_corr < .005), 120 minutes (t(1278) = 4.53, p_corr < .005) and 230 minutes (t(1073) = 5.28, p_corr < .001) post-infusion, and on days 1 (t(867) = 3.91, p_corr = .053), 2 (t(723) = 5.51, p_corr < .001) and 3 (t(614) = 3.79, p_corr = .087). The authors note degrees of freedom reported derive from SPSS mixed-model estimates. Secondary clinical analyses: There was no significant main effect of melancholic versus other depressive subtypes (F(1,51) = 2.001, p = .163) and no subtype-by-time interaction. Comorbid anxiety diagnosis did not predict differential anti-anhedonic response. A positive family history of alcohol use disorder in a first-degree relative was associated with an overall improved anti-anhedonic response (main effect F(1,54) = 4.538, p = .038), though the interaction with time was non-significant. PET imaging findings: PET was acquired in 20 patients (one excluded from ROI modelling due to missing venous sampling, yielding n = 19 for ROI analyses). ROI analyses showed no significant associations between baseline rCMRGlu in VS or OFC and baseline SHAPS, nor between baseline metabolism in these ROIs and subsequent SHAPS change. A trend-level negative correlation was found between increased VS rCMRGlu and decreased anhedonia (r(19) = -0.422, p = .072), which became non-significant when controlling for change in total MADRS (r(19) = -0.319, p = .197). Using small-volume correction centred on a previously reported dACC coordinate, increased dACC rCMRGlu following ketamine was significantly associated with decreased anhedonia (MNI X = -14, Y = 34, Z = 32, t(19) = 5.78, voxel-level p_SVC = .002); this remained significant after controlling for change in total MADRS (X = -14, Y = 34, Z = 30, t(19) = 5.15, p_SVC = .007). Whole-brain analyses revealed no baseline rCMRGlu predictors of SHAPS, but changes in metabolism correlated with change in anhedonia: increased glucose metabolism in the right hippocampus (extending into entorhinal cortex) was associated with decreased anhedonia (cluster-level p_WBC = .01), with a contralateral trend (p_WBC = .058). Decreased rCMRGlu in the right OFC (cluster-level p_WBC = .01) and left inferior frontal gyrus (cluster-level p_WBC = .045) were also associated with reduced anhedonia. When controlling for change in total MADRS (minus anhedonia item), the hippocampal and OFC relationships remained significant or at trend level, whereas the IFG relationship did not. The authors emphasise the exploratory nature of these imaging results and note limited correction for multiple comparisons.

Lally and colleagues interpret their re-analysis as showing a rapid and partially sustained reduction in anticipatory anhedonia following a single open-label subanaesthetic ketamine infusion in medication-free, treatment-refractory MDD patients, with measurable effects within 40 minutes and persisting up to 3 days. Adjunctive riluzole, started 4–6 hours post-infusion and continued for 28 days, did not enhance the anti-anhedonic response. In the PET subsample, decreased anhedonia was associated with increases in hippocampal and dACC glucose metabolism and decreases in OFC metabolism; several of these relationships held or trended after controlling for changes in overall depressive symptoms. The authors situate these findings relative to prior work: the dACC association replicates an earlier report in bipolar disorder, supporting a possible role for dACC metabolism in motivation and reward anticipation. The hippocampal (subicular) increases were unanticipated; Lally and colleagues suggest that heightened subiculum metabolism could plausibly influence mesolimbic dopamine release and thus reward-related processes, but they also note animal data showing non-linear subiculum–dopamine dynamics that may explain the absence of a robust ventral striatum metabolic finding. Decreased OFC metabolism is discussed in the context of reduced punishment or cost sensitivity during reward anticipation and decision making, consistent with known OFC involvement in value representation. The authors acknowledge several limitations that constrain causal inference and generalisability. Chief among these is the lack of a placebo-controlled ketamine condition, leaving open expectancy or non-specific effects; the treatment-refractory sample and prior randomised ketamine trials argue against a large placebo effect but cannot eliminate it. Analytical limitations include difficulties in modelling change in anhedonia independently of total depression score because baseline variance differed markedly from post-infusion time points, the timing mismatch between baseline PET (1–3 days pre-infusion) and baseline SHAPS (60 minutes pre-infusion), a small PET sample, and incomplete correction for multiple comparisons in imaging analyses. Acute psychotomimetic/dissociative effects at early post-infusion time points raise questions about the validity of the 40-minute anti-anhedonic signal, although prior data suggest larger acute dissociation may predict later antidepressant benefit. Finally, the authors concede that metabolic changes tied to anhedonia may reflect other unmeasured constructs (for example, sensitivity to environmental volatility) and that restricted baseline variance in a depressed sample may limit detection of baseline brain–behaviour relationships. In terms of implications, the paper stresses the potential of NMDA receptor antagonists to treat anhedonia and recommends future studies that include placebo-controlled designs, broader ranges of anhedonia severity, larger imaging samples, explicit tests of symptom-domain specificity and exploration of biomarkers (for example, family history of alcohol use disorder) to identify subgroups most likely to benefit.